The Confessions of a Protein Language Model: Steering, Activations, and What Predicted Confidence Cannot Certify
Protein generate-and-screen pipelines use interventions and readouts to select candidates, but rarely test whether these track their intended targets. We audit published repetition steering, ESMFold confidence, and an activation readout. Steering preserves confidence at its operating point, but a fitted cliff follows at alpha_50 = 1.240x (sweep's own control: 64.0%). Direction matters at moderate…
We haven't written up this one. bioRxiv has the full story — the link below goes straight to it.
